Connotation, characteristics and scale measurement of the intelligent economy
Intelligent Economy is a Chinese Original
Most of the professional terms we use were first proposed by foreigners. There are mainly 12 of them: Digital Economy, Artificial Intelligence, Big Data, Industrial Internet, Cloud Computing, Internet of Things, E-commerce, E-government, Platform Economy, Smart City, Unicorn, and Metaverse. The Intelligent Economy is different; it is a Chinese original. There are three key points regarding its proposal:
First, in 1990, three scholars from Guangxi published an influential article titled "Propositions for 21st - Century Economics: On the Intelligent Economy", making a forward - looking strategic judgment that the 21st century would be the era of the intelligent economy. This is the earliest relevant literature.
Second, in 2017, when Artificial Intelligence attracted the attention of countries around the world, the State Council's "New Generation Artificial Intelligence Development Plan" first proposed the term "Intelligent Economy" in an official document, calling for "forming an intelligent economic form driven by data, with human - machine collaboration, cross - border integration, and co - creation and sharing." The following year, the ninth group study session of the Political Bureau of the CPC Central Committee emphasized it again.
Third, this year's "Government Work Report". In 2025, the State Council's "Opinions on Deeply Implementing the 'Artificial Intelligence +' Initiative" listed the Intelligent Economy as one of the core development goals. This year's "Government Work Report" included the Intelligent Economy for the first time and gave it a separate paragraph, using the "new form of the Intelligent Economy" to guide all work. The government work reports of Sichuan, Jiangsu, Anhui, Guangxi, Liaoning, Chengdu, and Hangzhou also mentioned the Intelligent Economy. Guangxi listed the Intelligent Economy as its top priority this year.
Understanding Artificial Intelligence is the Key to Recognizing the Intelligent Economy
There are numerous definitions of Artificial Intelligence, and their core commonality is: a technology that completes various tasks by simulating human perception, learning, and behavioral abilities, which reflects the ability of machines to respond in a way consistent with human reactions. According to the definition of the OECD (2024), an Artificial Intelligence system is a machine - based system that makes inferences based on received inputs for explicit or implicit goals, generating outputs such as predictions, content, recommendations, or decisions, and thus having an impact on the physical or virtual environment.
There is no unified definition of the Intelligent Economy yet. Simply put, the Intelligent Economy refers to economic activities related to Artificial Intelligence, and it is an upgraded version of the Digital Economy. The Digital Economy is the "nerve" of the economic system, solving the problem of information interconnection; the Intelligent Economy is the "brain" of the economic system, with the abilities of autonomous perception, autonomous learning, autonomous decision - making, and continuous evolution.
The Intelligent Economy has only recently come into the public view, and there is no official English translation yet. In China, there are mainly two translations, "Intelligent Economy" and "Smart Economy", but neither seems to capture the essence. Only by translating "intelligent" as "AI" can the essential connotation of the Intelligent Economy be reflected and the scope be defined - "intelligent" cannot be clearly defined, but "AI" can. Therefore, translating the Intelligent Economy as "AI Economy" seems more appropriate.
Token is an Important Symbol of the New Form of the Intelligent Economy
Thirty years ago, the industrial economy was transitioning to the digital economy. Nicholas Negroponte's "Being Digital" emerged, providing Chinese people with digital enlightenment. The first sentence of the book is: "The best way to understand the value and impact of digital life is to think about the difference between bits and atoms." The core idea of the whole book is "Move bits, not atoms."
Today, as the digital economy is accelerating its transformation towards the intelligent economy, the role of Token in the intelligent economy is similar to that of atoms in the industrial economy and bits in the digital economy.
Token is the basic unit for large models to process and generate information, with the characteristics of measurability, pricing, and tradability. AI cannot directly understand human text, images, audio, and video. It first cuts our input into Tokens, then maps them to unique digital IDs. After calculation and processing, it converts them into the results we need and outputs them. Token belongs to the category of calculation and is the product of a segmentation strategy. It can be a word, a character, a punctuation mark, a piece of code, or a pixel block in an image or video. Token is often translated as "token", and it should be noted that it is not limited to language and text but also processes multi - modal information such as pictures and videos.
When AI evolves from a chat tool to an intelligent agent capable of autonomously performing tasks, the consumption of Tokens will increase exponentially. TPD (Token Per Day) has become a key indicator for measuring the activity and development level of large models.
Recently, the media has been filled with statements such as "China's large - model call volume exceeds that of the United States (Source: OpenRouter)" and "Token - related electricity export". These need to be carefully examined, and we cannot generalize. In March this year, China's TPD exceeded 140 trillion (Source: National Data Bureau). Google disclosed in October last year that its TPD reached 43 trillion. So far, the week starting on March 30 had the highest call volume on the OpenRouter platform, with a TPD of 3.86 trillion, which is only 9% of Google's six months ago; among them, China's TPD was 1.85 trillion, only 1.3% of China's total. OpenRouter's representativeness of the overall Token consumption is very limited, mainly reflecting the usage of individual developers and startup teams rather than large - scale enterprise deployments. Moreover, Chinese models do not equal local computing power in China. Among the 1.85 trillion TPD, only a small part of the computing power is deployed in China, and most calls are made to overseas nodes of Chinese models. The limited consumption of local electricity in China does not mean that China has achieved large - scale electricity export through Tokens.
The Next Measurement Frontier
We have entered the era of the intelligent economy, but the national economic industry classification has not made any definitions or accounting explanations for AI. Looking around the world, only the United Nations' "System of National Accounts" (2025 SNA) added an AI classification in its revision last year, classifying it as an independent type of computer software.
Fortunately, countries have begun to explore methods for measuring the intelligent economy. The U.S. Bureau of Economic Analysis (BEA) announced in July 2024 that it would no longer measure the scale of the digital economy and instead study methods for measuring AI output. The UK's Department for Science, Innovation and Technology (DSIT) has established a business - centered classification system, classifying AI from three levels: professionalism, business model, and capabilities, and has released multiple annual reports. According to calculations, there are 5,862 AI companies in the UK, with a total industry revenue of £23.9 billion and a gross value added (GVA) of £11.8 billion in 2024. According to the Ministry of Industry and Information Technology, in 2025, the number of AI companies in China exceeded 6,000, and the scale of the core industry is expected to exceed 1.2 trillion yuan. However, the calculation methods and scopes are kept secret.
Ten years ago, the digital economy was booming, and measuring the scale of the digital economy became a focus of attention for countries. The OECD, the U.S. BEA, the UK's DCMS, Statistics Canada, the Australian Bureau of Statistics, and China's National Bureau of Statistics have all constructed digital economy satellite accounts for scale measurement. Now, as the digital economy is becoming the mainstream of the economy and the AI revolution is surging, accurately measuring the scale of the intelligent economy is extremely urgent and has become the next measurement frontier (NBER, 2025). However, this work is still in its infancy, and relevant research is still relatively scarce. The times call on economists and statisticians to accelerate the construction of intelligent economy satellite accounts and measurement methods. As Huang Juechu et al. wrote in their 1990 paper on the intelligent economy: Engaging in this exploration does not require the huge equipment of a spaceship or a polar expedition, nor the thrilling experience of being in space or the polar regions. The research funds consumed are negligible compared to any new hardware development system, plus a little coffee and some writing paper, but what is obtained is a compass to guide the world economy to smoothly transition to the intelligent economy era.
Of course, measuring the intelligent economy is very difficult, mainly because AI is an enabling technology. Many great innovations in history have manifested as new products or services, such as cars, airplanes, telephones, and the Internet. Usually, a brand - new production and service system is created, rather than being integrated into the old system. AI is very different. Currently, it is still the superposition of new capabilities on the old structure. Academician Li Guojie (2025) pointed out that the essence of AI is an optimization technology to achieve expected goals. Currently, artificial intelligence is mainly an enabling technology, that is, a technology that adds icing on the cake from 1 to N. Although AI has also created new product forms, it mainly focuses on integration, empowerment, cost reduction, and efficiency improvement, and is integrated into existing products and services. For example, AI is widely used in digital advertising, online games, online meetings, customer service, and autonomous driving. This integration makes it impossible to accurately segment. This is not only reflected in the application layer but also in the underlying infrastructure. AI depends on industries such as semiconductors, data centers, and big data, which also meet other non - AI needs. They share these infrastructure resources. You are in me, and I am in you.
After all, AI is regarded as a revolution. If AI can be measured using existing methods and frameworks, then it is not a real revolution.
Yan Deli Senior Expert at Tencent
This article is from the WeChat official account "Tencent Research Institute" (ID: cyberlawrc). Author: Yan Deli. Republished by 36Kr with permission.